Fostering the Adoption of i by Practitioners: Some Challenges and Research Directions - Intentional Perspectives on Information Systems Engineering - page 182

Item
Selector

Item
Selector

Choose
Available
Item

Choose
Non-Available
Item

Item
Selection

Car Rental
Company
Walk-in
Rental

Walk-in
Customer

Rent a Car

(TD;|when;|whom);

|Task 5

Fig. 3 Structuring i ∗ models using new constructs: aspects [ 3] ( left ) and services [16] ( right )

identification of aspects and modularization of the model is made after the model
has been written, therefore the development process is still not recorded. Also, the
addition of aspects into i ∗ results in a framework with more modelling constructs
and may eventually require a steeper learning curve.
Estrada incorporates the concept of service into the i ∗ framework [ 16] (see Fig. 3,
right). This type of modularity unit is closer to the concepts managed in the domain
(i.e., business services) and from this point of view fits better than aspects to the
natural stepwise refinement process. However it is true that this particular proposal
introduces a lot of complexity to the framework, with the fundamental concepts
of “service” and “process”, and also with the configuration of services inside SR
boundaries using a variability-like model with mandatory and optional features
combined in several ways. As in the approach above, validation is needed to assess
usability of the proposal.
If we consider proposals aligning with stepwise refinement, we still find several
proposals. First, we mention the work by Leite et al. [ 40] that proposes a third kind
of i ∗ diagram to complement the SD and SR diagrams, namely the SA (Strategic
Actor) diagram that represents all kind of model actors (roles, positions and agents)
with their relationships (plays, occupies, covers, instance, is-a, is-part-of), see Fig. 4,
left, for an example. Also, Alencar et al. [ 1] introduce the concept of alternative for
grouping means to achieve an end. This concept is generalized by Franch [ 21] into
the general notion of module that is specialized into different types of SD- and
SR-modules (see Fig. 4, right, for example). Remarkably, these three approaches
define the introduced constructs in terms of metamodels.
Last, Lucena et al. [ 43] present a modularization approach that is built on
the existing framework without any extension. This means using actors as only
encapsulation mechanism. They provide a transformational framework in which
transformation rules convert logically connected subgraphs of existing SR diagrams
into SR diagrams for new actors.